ReplyZen sends DMs from two different starting points:
- The DMs inbox (this section). For an inbound message not owned by an automation, the agent follows its visible Off, Draft, or Automatic setting.
- Automations (automations docs). A comment matches a trigger condition, and the automation sends a follow-up DM to that user as part of its action.
Both end up in the same conversation thread. This page covers how they interact.
Where the conversation lives
There's only ever one thread per user per page. If an automation sent a DM to a user yesterday and the user replies today, the reply lands in the same inbox thread you'd see if they had messaged out of the blue. You don't have to hunt across two screens to find context.
The thread view shows the full history: the automation's opening message, any 2-step consent flow, the user's replies, every AI draft and human send that followed.
"Triggered by a comment"
If a conversation started as an automation reaction to a comment, the thread header shows a small Triggered by a comment link. Clicking it opens the original comment in the Comments inbox. Useful when you want to see what the user said publicly before they ended up in your DMs.
The link is only present when the conversation actually started from an automation. Free-text inbound DMs don't have one.
One owner per inbound message
When an explicit automation matches an inbound message, that automation owns the turn and uses its own publish behavior. Inbox handling does not generate a second reply. If no automation matches, the agent's Inbox handling setting applies.
If the automation matches but its visible conversation guard blocks it, the automation still owns that turn and ReplyZen stays silent. This is how a once-only automation avoids accidentally handing later messages to automatic inbox handling.
Handoff stops both AI paths
When you click I'll take it from here on a conversation, the AI inbox stops drafting and any automation that would normally send a DM in that thread is blocked. The handoff is respected across both systems.
This matters for a specific scenario: a user comments on a post, an automation is configured to send them a follow-up DM, but in the meantime the user has already DM'd you about the same thing and you've taken the conversation over manually. Without the handoff guard, the automation would send its scripted DM on top of your in-progress human conversation. With it, the automation silently skips.
If you later Reopen for AI, both paths come back online. New automation triggers can send again, and the inbox follows its configured mode.
Configuring automation DMs
The automation-side configuration (what the message says, single message vs 2-step consent, button links, skip-if-conversation-exists) is in the Setting up an automation doc. Nothing about that has changed; the inbox just observes the outputs.
When to use which
A short reference:
| Situation | Right tool |
|---|---|
| User comments on a post asking about returns | Automation that replies publicly and DMs return instructions |
| User comments saying "DM me the link" | Automation with a 2-step consent DM |
| User messages your page directly and no automation matches | Inbox handling: Off, Draft, or Automatic |
| User message matches an explicit automation | That automation alone owns the turn and follows its publish behavior |
| User replies to an automation-initiated DM and no new automation matches | Inbox handling, in the same thread the automation started |
| You need to walk a user through a multi-turn problem | Inbox, with handoff to manual when the AI runs out of useful answers |